Python Developer | AI/ML Engineer | Building intelligent systems that run autonomously
- Software Developer at Nio Stars Technologies, working on AI/ML and backend systems
- I build Python applications that combine machine learning with production engineering — from NLP pipelines and computer vision models to autonomous agents and automated research systems
- Currently focused on multi-agent orchestration, local LLM inference with Ollama, and automated security intelligence
Languages & Frameworks
AI / ML
Tools & Infrastructure
| Project | What It Does | Tech |
|---|---|---|
| ai-research-agent | Searches GitHub daily for new AI repos, analyzes READMEs & tech stacks, generates LLM-powered summaries and comparative reports | Python, Ollama, GitHub API, GitPython |
| daily-experiments | Pulls real CVEs from NIST NVD, computes threat indices, generates severity dashboards with day-over-day delta tracking | Python, NIST NVD API, Matplotlib |
| automated-bots | Algorithmic market signal generator — SMA crossover, RSI, momentum analysis with 4-panel visual dashboards | Python, Matplotlib |
| ml-experiments | Hyperparameter tuning across 4 ML tasks (churn, fraud, pricing, demand) with loss curves and feature importance tracking | Python, Matplotlib |
| agent-improvement | Perceive->Reason->Act->Refine decision loop across operational domains with risk scoring and self-correction | Python, Matplotlib |
| ai-agent-lab | Multi-agent pipeline simulator (planner->researcher->coder->reviewer->synthesizer) tracking quality, latency, and token usage | Python, Matplotlib |
| Project | What It Does | Tech |
|---|---|---|
| RAGify-Finance | Benchmarks Cohere vs HuggingFace embeddings for financial document Q&A using RAG | Python, LangChain, FAISS, Cohere, Gemini |
| AI_Invoice_Master | Multi-language invoice extractor — OCR + Gemini AI for intelligent data extraction | Python, Tesseract, Streamlit, Gemini |
| DocuMind | PDF document comparison with AI-powered change detection and analysis | Python, LangChain, Gemini |
| GemmaQA-DocExplorer | RAG-powered PDF Q&A using Groq Gemma2-9b and FAISS vector search | Python, Groq, FAISS, Streamlit |
| TalentGPT | Multilingual AI recruiter — conducts technical screening with sentiment analysis | Python, Streamlit, LLM |
| Task-Extraction-NLP | Extracts tasks from unstructured text, identifies person/action/deadline using NLP | Python, NLTK, Word2Vec |
| Project | What It Does | Tech |
|---|---|---|
| pneumonia-xray-classification | Detects pneumonia from chest X-ray images using CNNs | Python, TensorFlow/Keras, CNN |
| CNN-Emotion-Detection | Classifies 7 facial emotions from FER-2013 dataset (48x48 grayscale) | Python, Keras, OpenCV |
| ai-video-recommendation-system | Video recommendation engine with REST API | Python, FastAPI, SQL, ML |
| Project | What It Does | Tech |
|---|---|---|
| google-meet-transcriber | Captures and saves real-time transcript data from Google Meet | Rust |
| Advanced_Gemini_QA_Bot | Interactive Q&A chatbot with chat history using Gemini Pro | Python, Streamlit, Gemini |
- Building automated research agents that discover, analyze, and report on AI repositories daily using local LLMs
- Security intelligence pipeline pulling real-world CVE data from NIST and tracking threat trends
- Multi-agent orchestration patterns — modeling how specialized agents collaborate on complex tasks